Data Analysis - Velib Project in Python  ¶

import pandas as pd

import numpy as np

import random as rd

import matplotlib.pyplot as plt

import seaborn as sns

sns.set_style('darkgrid')

​

%matplotlib inline

### TO BE COMPLETED ### 

loading = pd.read_csv('data/velibLoading.csv')

# %load solutions/Python/load_loading.py

path    = 'data/'  # If data in 'data' directory

loading = pd.read_csv(path + 'velibLoading.csv', sep = " ")

​

loading

### TO BE COMPLETED ### 

coord = np.pd.read_csv('data/velibCoord.csv')

# %load solutions/Python/load_coord.py

coord = pd.read_csv(path + 'velibCoord.csv', sep = " ")

​

coord.head()

### TO BE COMPLETED ### 

​

print(coord.isnull().sum())

#aucune donnée manquante = aucune donnée non renseignée

​

print(loading.isnull().sum().sum())

#la premiere sum regroupe les valeurs de chaque colonne et la deuxieme regroupe toutes les valeurs du tableau

# %load solutions/Python/missing_value.py

loading_missing_value = loading.isna().sum().sort_values(ascending=False)

​

print('--- Loading ---')

print(loading_missing_value.sum())

​

# --- #

print('')

​

coord_missing_value = coord.isna().sum().sort_values(ascending=False)

​

print('--- Coord ---')

print(coord_missing_value)

### TO BE COMPLETED ### 

​

​

print('--- Loading ---')

print(loading.duplicated().sum())

​

# --- #

print('')

​

print('--- Coord ---')

print(coord.duplicated().sum())

# %load solutions/Python/duplicated.py

print('--- Loading ---')

print(loading.duplicated().sum())

​

# --- #

print('')

​

print('--- Coord ---')

print(coord.duplicated().sum())

### TO BE COMPLETED ### 

​

# Stations in descending order of occurrence

station_name = coord['names']

print(station_name)

​

station_name.value_counts()

# %load solutions/Python/station_name.py

# Stations in descending order of occurrence

station_names = coord.names.value_counts().sort_values(ascending=False)

print(station_names)

​

# --- #

print('')

​

# We display the station with the most occurrences, i.e. the station corresponding to the first line of 'station_name'.

name = station_names.index[0]

coord[coord.names == name]

### TO BE COMPLETED ### 

​

i = rd.randrange(loading.shape[0])

loading_data = loading.to_numpy()

​

n_steps = loading.shape[1]  # number of observed time steps

time    = np.linspace(1,n_steps,n_steps)  # observed time range

time_tick  = np.linspace(1, n_steps, 8)  # beginning of days

​

# --- #

​

plt.figure(figsize = (20, 6))

​

plt.plot(time,loading_data[i, :])

plt.vlines(x = time_tick, ymin = 0, ymax = 1, 

           colors = "orange", linestyle = "dotted", linewidth = 3)

​

plt.xlabel('Time', fontsize = 20)

plt.ylabel('Loading', fontsize = 20)

plt.title(coord.names[1 + i], fontsize = 25)

plt.show()

# %load solutions/Python/plot_loading.py

i = rd.randrange(loading.shape[0])

loading_data = loading.to_numpy()

​

n_steps    = loading.shape[1]  # number of observed time steps

time_range = np.linspace(1, n_steps, n_steps)  # observed time range

time_tick  = np.linspace(1, n_steps, 8)  # beginning of days

​

# --- #

​

plt.figure(figsize = (20, 6))

​

plt.plot(time_range, loading_data[i, :], linewidth = 2, color = 'purple')

plt.vlines(x = time_tick, ymin = 0, ymax = 1, 

           colors = "orange", linestyle = "dotted", linewidth = 3)

​

plt.xlabel('Time', fontsize = 20)

plt.ylabel('Loading', fontsize = 20)

plt.title(coord.names[1 + i], fontsize = 25)

plt.xticks(fontsize = 15)

plt.yticks(fontsize = 15)

plt.tight_layout()

plt.show()

### TO BE COMPLETED ### 

​

fig,axs =plt.subplots(4,4,figsize = (15,12))

​

​

​

for i in range(4):

    for j in range(4):

        k=stations[2 * i + j]

        axs[i,j].plot(time,loading_data[k, :])

        axs[i,j].vlines(x = time_tick, ymin = 0, ymax = 1, colors = "orange", linestyle = "dotted", linewidth = 3)

        axs[i, j].set_title(coord.names[1 + k], fontsize = 12)

# %load solutions/Python/plot_loading_16.py

# We select 16 stations at random

stations = np.arange(loading.shape[0])

rd.shuffle(stations)

stations = stations[:16] 

​

# --- #

​

fig, axs = plt.subplots(4, 4, figsize = (15,12))

​

for i in range(4):

    for j in range(4):

        k_station = stations[4 * i + j]

        axs[i, j].plot(time_range, loading_data[k_station, :], linewidth = 1, color = 'purple')

        axs[i, j].vlines(x = time_tick, ymin = 0, ymax = 1, colors = "orange", linestyle = "dotted", linewidth = 3)

        axs[i, j].set_title(coord.names[1 + k_station], fontsize = 12)

​

for ax in axs.flat:

    ax.set_xlabel('Time', fontsize = 12)

    ax.set_ylabel('Loading', fontsize = 12)

    ax.tick_params(axis='x', labelsize=10)

    ax.tick_params(axis='y', labelsize=10)

plt.tight_layout()

plt.show()

### TO BE COMPLETED ### 

​

bp=plt.boxplot(loading, patch_artist = True)

# %load solutions/Python/plot_loading_disp.py

plt.figure(figsize = (20,6))

​

# --- #

​

bp = plt.boxplot(loading_data, widths = 0.75, patch_artist = True)

​

for box in bp['boxes']:

    box.set_alpha(0.8)

for median in bp['medians']:

    median.set(color = "Purple", linewidth=5)

# --- #

plt.vlines(x = time_tick, ymin = 0, ymax = 1, 

           colors = "Orange", linestyle = "dotted", linewidth = 5)

​

# --- #

​

plt.xlabel('Time', fontsize = 20)

plt.ylabel('Loading', fontsize = 20)

plt.title("Boxplots", fontsize = 25)

plt.xticks(ticks = np.arange(0, 168, 5), labels=np.arange(0, 168, 5), fontsize = 15)

plt.yticks(fontsize = 15)

​

plt.tight_layout()

plt.show()

### TO BE COMPLETED ### 

​

print('--- Average fill rate ---')

print(loading.mean().mean())

#On veut la moyenne de toutes les stations à toutes les heures

​

# Pour avoir la moyenne des colonnes pour une ligne en particulier (une station) : print(loading.mean(axis=1))

​

# --- #

print('')

​

print('--- Least crowded station, on average ---')

mean=loading.mean(axis=1)

i = mean.idxmin()

print('Average fill rate :',mean[i])

print(coord.loc[i])

​

​

# --- #

print('')

​

print('--- Fullest station, on average ---')

mean=loading.mean(axis=1)

i = mean.idxmax()

print('Average fill rate :',mean[i])

print(coord.loc[i])

​

[...]

​

# %load solutions/Python/loading_mean.py

print('--- Average fill rate ---')

print(loading.mean().mean())

​

# --- #

print('')

​

loading_mean = pd.Series(loading.mean(axis=1))

​

print('--- Least crowded station, on average ---')

i = loading_mean.idxmin()

print('Average fill rate :',loading_mean[i])

print(coord.loc[i])

​

# --- #

print('')

​

print('--- Fullest station, on average ---')

i = pd.Series(loading.mean(axis=1)).idxmax()

print('Average fill rate :',loading.mean(axis=1)[i])

print(coord.loc[i])

### TO BE COMPLETED ### 

​

​

plt.plot(mean)

plt.hlines(y=loading.mean().mean(),xmin=0,xmax=1189,colors='orange',linewidth = 5)

​

# %load solutions/Python/plot_mean_stations.py

n_stations = loading.shape[0]  # number of observed stations

stations   = np.arange(n_stations)

​

plt.figure(figsize = (20,6))

​

# --- #

​

plt.plot(loading_mean)

plt.hlines(y = loading.mean().mean(), xmin=0, xmax=n_stations, 

           colors = "OrangeRed", linewidth = 3)

​

# --- #

​

plt.xlabel('Stations', fontsize = 20)

plt.ylabel('Average loading', fontsize = 20)

plt.title("Average loading per station", fontsize = 25)

plt.xticks(fontsize = 15)

plt.yticks(fontsize = 15)

​

plt.tight_layout()

plt.show()

### TO BE COMPLETED ### 

​

mean_per_hour_per_day = loading.mean(axis = 0).to_numpy()

mean_per_hour_per_day = mean_per_hour_per_day.reshape((7, 24))

days = ["Monday", "Tuesday", "Wednesday","Thursday", "Friday", "Saturday", "Sunday"]

plt.figure(figsize = (15,10))

plt.plot(mean_per_hour_per_day.transpose())

plt.plot(mean_per_hour, color = "black", linewidth = 3)

# %load solutions/Python/plot_mean_hours.py

mean_per_hour_per_day = loading.mean(axis = 0).to_numpy()

mean_per_hour_per_day = mean_per_hour_per_day.reshape((7, 24))

​

mean_per_hour = mean_per_hour_per_day.mean(axis=0)

​

# --- #

​

days = ["Monday", "Tuesday", "Wednesday","Thursday", "Friday", "Saturday", "Sunday"]

plt.figure(figsize = (15,10))

​

plt.plot(mean_per_hour_per_day.transpose())

plt.plot(mean_per_hour, color = "black", linewidth = 3)

​

plt.xlabel('Hourly loading, averaged over all stations', fontsize = 20)

plt.ylabel('Loading', fontsize = 20)

plt.legend(days + ['Weekly'])

plt.xticks(ticks = np.arange(0,24,4), labels=np.arange(0,24,4), fontsize = 15)

plt.tight_layout

plt.show()

import matplotlib.cm as cm

import matplotlib.patches as mpatches

import plotly.express as px

### TO BE COMPLETED ### 

## Simple 2D representation

# Monday at hour 6h, 12h, 23h

​

# Hours to be displayed

hours = ...

​

# --- #

​

[...]

# %load solutions/Python/plot_loading_2D_monday.py

## Simple 2D representation

# Monday at 6h, 12h and 23h

​

hours = [6, 12, 23]

​

# --- #

​

s, n = 10, len(hours)

fig, axs = plt.subplots(1, n, figsize = (s*n, s))

​

for (i,h) in enumerate(hours):

    im = axs[i].scatter(coord.latitude, coord.longitude, c = loading_data[:,h], cmap = cm.plasma_r)

    axs[i].set_title('Stations loading - Monday {} h'.format(h), fontsize = 25)

    plt.colorbar(im, ax=axs[i])

for ax in axs.flat:

    ax.set_xlabel('Latitude', fontsize = 20)

    ax.set_ylabel('Longitude', fontsize = 20)

    ax.tick_params(axis='x', labelsize=15)

    ax.tick_params(axis='y', labelsize=15)

​

plt.tight_layout()

plt.show()

### TO BE COMPLETED ### 

## Simple 2D representation

# Loading at 6pm, depending on the day of the week

​

[...]

# %load solutions/Python/plot_loading_2D_18h.py

## Simple 2D representation

# Loading at 6pm, depending on the day of the week

​

h = 18

hours = np.arange(h, 168, 24)

​

load_per_hour = loading_data[:, hours]

​

days = ["Monday", "Tuesday", "Wednesday", "Thursday", "Friday", "Saturday", "Sunday"]

​

# --- #

​

s, m = 10, 3

k = 1 + len(days)//m

​

fig = plt.figure(figsize=(s+1, s))

fig.subplots_adjust(left=0, right=1, bottom=0, top=1, hspace=.3, wspace=.25)

​

for (i,d) in enumerate(days):

    ax = fig.add_subplot(k, m, i+1)

    im = ax.scatter(coord.latitude, coord.longitude, c = load_per_hour[:,i], cmap = cm.plasma_r)

    plt.colorbar(im)

    ax.set_title('Stations loading - ' + d + ' {} h'.format(h))

    ax.set_xlabel('Latitude')

    ax.set_ylabel('Longitude')

    ax.tick_params(axis='x')

    ax.tick_params(axis='y')

​

#plt.tight_layout()

plt.show()

### TO BE COMPLETED ### 

## Visualization on the Paris map

​

[...]

loading_hill = ...

​

[...]

# %load solutions/Python/hilltop_stations.py

### TO BE COMPLETED ### 

## Simple 2D representation

​

[...]

# %load solutions/Python/hilltop_stations_2D.py

### TO BE COMPLETED ### 

## Visualization on the Paris map

​

coord['hill'] = coord['bonus'].astype('category') # convert to categorical

​

[...]

# %load solutions/Python/hilltop_stations_map.py

​

​

​

​